Immunotyping and Quantification of Melanoma Tumor–Infiltrating Lymphocytes

Summary/Abstract:

The density of tumour-infiltrating lymphocytes (TILs) in melanoma is correlated with improved clinical prognosis; however, standardized TIL immunotyping and quantification protocols are lacking. Herein, we provide a review of the technologies being utilized for the immunotyping and quantification of melanoma TILs.

1. Introduction

The density of tumour-infiltrating lymphocytes (TILs) is associated with improved prognosis in numerous cancer types, including metastatic melanoma tumors [1–6]. TIL density is believed to be indicative of an anti-tumor immunologic response. At this time, the TNM staging system for melanoma, which incorportates Breslow (tumor) thickness, presence of ulceration, mitotic rate, and lymph node and distant metastasis, is the main determinant of tumor prognosis, and therapy [7].

Recent advances in immunotherapy have greatly improved survival in metastatic melanoma [8,9], but response to immunotherapies is variable. It is unknown why melanomas of the same stage can have variable therapeutic responses, but TIL density has been correlated with response to immunotherapy and may therefore be one reason for the variability in response [10–13]. For this reason, reporting of TILs has become routine in melanoma histopathology [14], and TILs remain a focus of investigation by melanoma researchers. Metastatic melanoma TIL density and pattern have been under increased scrutiny by both clinicians and researchers for prognostication and selection of immunotherapies [15]. Adapting from the TIL quantification experience in invasive breast cancer, initial guidelines have begun to address standardization of these techniques in melanoma [16–18]. Herein, we discuss the methods and relevant updates to immunotyping and quantifying melanoma TILs.

2. Tumor Sample Preparation and Tissue Microarray

Assessments of TILs are most commonly performed by immunohistochemical (IHC) methods which involve taking sections of tumor. As tumors are three-dimensional, taking whole-tumor sections, staining, and processing are cumbersome and resource-/time-intensive. In an effort to save time and resources including the amount of reagents and tissue required for processing, Kononen et al [19] introduced the tissue microarray (TMA) in 1998. Tissue microarrays allow researchers to select tissue regions of interest, sample them by taking cores, and place multiple of these cores into parrafin blocks in parallel for subsequent sectioning and staining (Fig. 1). This allows for staining and processing of just one section at a time, but of multiple tumors or regions of interest concurrently [20]. TMAs may then be re-sectioned and stained for other markers of interest. Sections may be taken until the end of the core is reached.

Fig. 1.

Overview of tissue microarray construction. Donor tissue is placed in a FFPE block and sectioned. H&E staining is performed and regions of interest are marked. These regions are then sampled from the donor block using a tissue arrayer or hollow biopsy needle. The resulting cores are then placed in a recipient block which can be sectioned repeatedly for staining of the regions of interest. TMA sections can be stained for many types of analyses.

TMA construction for TIL assessment begins with selection of tumor regions of interest. Choosing appropriate tumor areas is imperative in order to perform a TIL assessment that is representative of the tumor or specific to the research question of interest. When performing TIL assessment, the authors suggest excluding regions of tumor with hemorrhagic or necrotic zones, ulcerations, tumor zones with crush artifacts or regressive hyalinization, and previous biopsy sites. If the research question regards the tumor margin, selecting these regions may be appropriate, but if not, then regions outside of the tumor borders should be avoided. Localization of TIL around blood vessels has been shown to be prognostic and may be of interest, but very large blood vessels should be excluded [21]. If no complete tumor specimen is available for analysis, biopsies can be used in the pre-therapeutic neoadjuvant setting.

Once tumor regions are selected, cores of those regions are taken and placed in parrafin blocks. Tissue cores for TMA construction can range from 0.6 mm to 2.0 mm [22,23]. Selecting the correct core size has significant implications, as larger core size captures more of the region of interest but limits the total number of samples per TMA block. The authors recommend the use of 1.0 mm diameter tissue cores from representative tumor areas from the FFPE tissue blocks, as this gives a large enough core for TIL assessment and allows for numerous cores per block.

It is important to include known tissue controls on each TMA block. Relevant control tissues such as from the spleen, liver, placenta, tonsil, and kidney should be included. This allows for staining of known cell types within these tissues as positive and negative controls. Multiple 4 μm thick sections can then be cut for Hematoxylin and Eosin (H&E) or IHC staining. After sectioning of the TMA block, the sections can then be stained with antibodies to the appropriate TILs. Common antibodies include CD3, CD4, CD8, CD20, CD34, CD45, CD56, CD138, CD163, DC-LAMP, FOXP3, and PD-1.

Once stained, the sections can then be reviewed for pattern, type, density, and location of tumor-infiltrating immune cells, tumor cell type, necrosis, haemorrhage, and melanin pigment. Properties of the sample, such as age, method of preservation, storage conditions, and tissue type, are important factors to consider in order to avoid antigenic deterioration and obtain high quality TIL immunotyping and quantification.

Ultimately, TMAs allow for repeated sections to be taken of the same region of interest within one or multiple tumors. These sections can then be stained with a wide array of antibodies for IHC or other analyses.

3. TIL Analysis and Quantification

With the introduction of immunotherapy, assessment of TILs and the immune environment within tumors is becoming ever more important. To characterize the cell populations within tumors and understand their interactions and relative spatial arrangements, it is necessary to be able to identify multiple cellular markers at once. A significant limitation of traditional immunohistochemistry (IHC) and flow-cytometry in performing these assessments is the inability to stain the same slide or sample for more than two targets at once. Several technologies have been developed which now allow for detection of many targets in one sample or slide of tissue simultaneously. These methods fall into two major categories: microscopy imaging based techniques such as multiplex immunofluorescence histology, and non-imaging based techniques such as time-of-flight mass cytometry (CyTOF). Here, we provide a brief overview of these technologies.

3.1. Multiplex IHC Analysis of TIL

Historically, IHC has been the most commonly used method to evaluate in-situ protein expression in a tumor sample. It detects the presence of an antigen on FFPE tissue through the use of primary monoclonal antibodies, enzyme-linked secondary antibodies, and precipitation reactions resulting in colorimetric chromogen deposition. While it is very commonly used, this method is significantly limited by the fact that congruent assessment of many proteins on the same subcellular region is limited. Surface co-expression or nuclear co-expression is difficult to assess using this method. Multiplex IHC (mIHC) allows for the visualization of multiple target proteins and the determination of their spatial arrangement within a single tissue section (Fig. 2). Multiplex IHC has been validated in multiple studies and can provide accurate quantitation of TIL infiltrates in IHC stained sections for melanoma [24–26].

Fig. 2.

Multiplex immunofluorescence histology image of a melanoma metastasis stained for identification of immune infiltrates and tertiary lymphoid structures (TLS). (a) 5-color multispectral composite image. (b) CD20 alone. (c) CD8 alone. (d) Ki67 alone. (e) Peripheral node addressin (PNAD) alone. (f) repeat of the 5-color multispectral composite image.

Multiple approaches to mIHC have been developed, and their differences center on how the staining and imaging steps are performed, though it is typically best done using fluorescent labels instead of chromogen deposition such as in traditional IHC. Generally, multiplexing is achieved by sequential staining of the tissue specimen of interest with antibodies to the target proteins. Each antibody is labelled with a specific fluorophore that fluoresces at a specific wavelength, thereby allowing for imaging of the target biomarker when exposed to its specific wavelength (Fig. 3a). Numerous techniques for staining and image capture have been developed, and the three general techniques for multiplexing are stain removal approaches, fluorophore inactivation, and DNA barcoding.

Fig. 3.

Methods of TIL Analysis and Quantification. (a) Multiplex immunofluorescence histology begins by creating tissue slides. This can be from a section of tissue or from a tissue microarray (TMA). The slide then undergoes sequential staining, antibody stripping, and imaging. Single-color images can then be merged. (b) Tyramide signal amplification increases the detection of low-concentration epitopes by a tyramide-tyrosine binding reaction. The slide is stained using tyramide-bound fluorophores and is then washed with hydrogen peroxide, which activates the horseradish peroxidase (HRP) bound to the secondary antibody. The HRP activates tyramide and allows for tyramide-tyrosine binding on the antibody-stained epitope. The antibodies are then stripped from the slide and the process can be repeated with the next antibody. (c) Time of flight mass cytometry (CyTOF) begins with tumor or tissue dissociation. The cells are suspended and passed through a nebulizer which creates single-cell vapor droplets. These are then burned and pass through a plasma torch which results in metal ion clouds, each from a single cell. A quadrupole filters the ion cloud before the ions enter the time of flight mass cytometer where the presence and quantity of each metal isotope is measured. This allows for a single-cell quantification of the antibody-labeled proteins.

Stain removal techniques (or “dye cycling”) utilize many iterations of staining, image capture, removal of stain, and re-staining. This allows for many stains to be applied and can result in a high number of markers to be studied in a single section of tissue. Two common methods for stain removal include multiepitope-ligand cartography (MELC) [27,28] and sequential immunoperoxidase labelling and erasing (SIMPLE) [29]. Fluorophore inactivation techniques are similar to stain removal, but focus instead on inactivating the fluorophores themselves via oxidation, denaturation, or photobleaching. While fluorophore inactivation and stain removal techniques do allow for analysis of a large number of targets, there are some limitations. For example, repeated staining and removal or bleaching of tissues may change the antibody binding affinity over time or result in tissue damage during processing [30]. Another significant downside of these techniques is that the primary antibody incubation time can be slow and with multiple rounds of labelling necessary, experiments can take up to a few days to perform.

DNA barcoding is another method for mIHC, and is relatively fast compared to the cycling techniques. This method works by applying many different antibodies labelled with DNA barcodes simultaneously. This is then followed by rapid barcode readout utilizing either fluorescing imager strands (DNA-Exchange, DEI) [31], or a process called Co-detection by indexing (CODEX) [32].

One of the challenges of mIHC in TIL analysis is that many target proteins of interest can be present at very low levels in the tumor microenvironment. As such, signal amplification techniques are needed for detection of proteins which occur at low-levels within tissues. Tyramide signal amplification, or TSA, significantly improves the detection of epitopes present in low levels and is commonly used when assessing TIL in melanoma [33–36]. TSA is an enzyme-mediated process in which an enzyme such as horseradish peroxidase (HSP) catalyzes a covalent binding reaction between tyramine-labelled fluorophores and the target protein (Fig. 3b). This process can deposit a considerable amount of fluorophore on the target and therefore results in significant signal amplification [24]. (Fig. 3 near here)

The need for accurate and reproducible description and quantification of TILs has resulted in the development of multiplex imaging and analysis platforms. To that end, semi-automated and automated machine-learning systems have become the standard for image analysis given the huge number of images that can now be taken. Vectra^®-Polaris^™ (PerkinElmer), an automated quantitative multispectral microscope, and inForm^® software have been validated in IHC and IF samples [37,38]. However, there are various software platforms available that can be used for the analysis of multispectral images from Vectra^®-Polaris^™ scanner systems, such as HALO^™ (Akoya Biosciences) [37]. Important considerations for analysis software are easy accessibility, automated capabilities of detecting tissue segmentation, compartmentalization of the staining, and spatial colocalization of cell distribution. With the use of digital quantification software, objective TIL quantification can be performed.

Despite its significant conveniences, mIHC has some limitations. One of these is the two-dimensional nature of stained slides in evaluating a three-dimensional tumor. This limitation may be managed by sampling different areas of interest within a tumor and constructing a TMA. False-positive labeling can take place due to cross-reactivity among different components of the reaction. This may occur during sequential labeling when the second round of antibodies intended for the second antigen bind the first antigen through the primary antibody. The effects of cross-reactivity are typically mitigated by including a positive and negative control tissue for each target of interest during panel development.

3.2. Flow Cytometric Analysis of TIL from Tumor Cell Suspensions

TIL can also be evaluated by non-imaging based methods. These typically involve flow cytometry (FC) or related techniques. FC has been used to evaluate melanoma TIL [39–42] but is limited by the number of fluorophores that can be used in one assay. Furthermore, FC is limited by its inability to render single-cell analyses. Time-of-flight mass cytometry is a technique similar to FC which allows for single cell analyses and analysis of protein expression levels.

3.3. Time-of-Flight Mass Cytometry (CyTOF) Analysis of TIL

The basic principle of CyTOF is similar to conventional FC-based analysis. Both modalities use antibodies to label their antigen targets. While FC antibodies are fluorophore labeled, CyTOF antibodies are conjugated to rare heavy metal isotopes (Fig. 3c). Due to the limited range and overlap of wavelengths emitted as a result of fluorophore excitation, FC is limited in the number of independently assessable parameters and thus requires large sample sizes to process multiple biomarkers.

CyTOF uses an atomic mass cytometer to identify and quantify the rare heavy metal isotope-labeled antigen [43]. Detection overlap among heavy metal isotopes is less than 2%, whereas spectral overlap seen in FC ranges from 5–100% [44]. Moreover, unlike FC samples which have to be processed within a few hours due to the use of fluorescent dyes, CyTOF metal-tagged samples can be cryopreserved for up to 4 weeks without notable loss of signal [45]. This can be especially helpful in instances of sampling of large-scale batches or equipment breakdown. CyTOF has successfully been used to analyze TIL from melanoma patients undergoing immunotherapy [46–51]. The high purity of the metal isotopes which are not found in normal tissue reduces background noise, eliminating spectral spillover, and avoids cellular autofluorescence associated with conventional FC [52]. One of the most significant benefits of CyTOF is the ability to not only identify proteins by labeling, but to quantify them as well. Furthermore, because the mass cytometer measures the isotope-labeled target protein in a single-cell suspension, the analysis results in single-cell resolution.

CyTOF has a number of disadvantages. Namely, the dimensional relationship between cells is lost as this is a suspension based technique. The use of isotopic metals in CyTOF are primarily from the lanthanides series. A panel of 40 antibodies can be used simultaneously to detect various biomarkers. Further research is being conducted to develop use of metals outside the lanthanides series to increase the number of parameters able to be processed simultaneously. As a result of CyTOF’s high number of parameters, data analysis is complex which has warranted new bioinformatics approaches to interpret and visualize the data [52]. Another disadvantage of CyTOF is its relatively high cost of metal tagged-antibodies, antibody conjugation kits, and reagents. With increasing demand and advancement of instrument manufacturing technology, CyTOF will likely become more accessible and be more affordable for a wide range of laboratories.

4. TIL Immunotyping

Immunotype classification describes the infiltration patterns, absolute cell counts, and densities of TILs. Immunotypes range from classification schemes such as dichotomization of infiltrates as ‘sparse versus dense, or “cold versus hot” [53], to more complex paradigms which incorporate both immune infiltrates and the tumor’s histologic architecture. Advantages of classifying tumors by immunotype include easy translation to the realm of clinical pathology, and potentially, better characterization of spatial relationships between immune cells and vasculature, tertiary lymphoid structures, and other tumor-associated histologic features.

The first comprehensive classification of TIL in melanoma was by Clark et al in 1989 and was done in primary tumors [54]. Lymphocytic infiltration patterns were stratified as brisk, non-brisk, or absent. To date, the Clark classification of TILs remains widely used in melanoma. This is due to its relative reproducibility [55] and wide validation as an independent prognostic factor when applied to primary cutaneous melanoma lesions and lymph node metastasis [56–58]. In 2012, the Melanoma Institute Australia (MIA) proposed an updated TIL classification system [5]. TILs were stratified from grade 0 through grade 3, based on immune infiltrate density (absent, mild, moderate, or marked) and distribution (absent, focal, multifocal, or diffuse) resulting in four TIL grades (Table 1). The MIA classification was found to be an independent predictor of melanoma-specific survival, but external validation studies are lacking [5].

Table 1.

Comparison of Melanoma Institute of Australia (MIA) and Clark TIL Classification

TIL Pattern	MIA	Clark
Absent of TILs	Grade 0	Absent
Mild/moderate focal infiltrate OR Mild multifocal infiltrate	Grade 1	Non-brisk
Marked focal/moderate/marked multifocal OR Mild diffuse infiltrate	Grade 2	Brisk
Moderate/marked diffuse infiltrate	Grade 3	Brisk

The authors have previously characterized the immunotype of 183 metastatic melanomas resected from 147 patients [21]. TILs were classified as immunotype A, B, or C if they were found to have no lymphoid infiltrate, perivascular infiltrates only, or diffuse intra-tumoral infiltrate, respectively. Following classification, lesions were assigned a numeric score based on immunotype. This immunotype classification and higher CD8+ T cell densities were associated with increased overall patient survival. This immunotype classification has been extended to a murine melanoma model [33] and found to correlate with PD-L1 expression by melanoma cells [59].

In 2017, an international immuno-oncology biomarkers working group proposed a standardized method for assessing TILs in all solid tumors [18,17]. Guidelines regarding TIL quantification include:

1. The reporting of TILs as a continuous variable (i.e. percentage of the stromal area) for ease of understanding and statistical analysis.

2. The separate reporting of TIL stromal and tumor cell compartment due to the difference in TIL density between the two compartments.

3. The evaluation and separate reporting of TILs within the borders of the invasive margin, defined as a 1 mm zone between the malignant cells and host tissue, and the rest of the tumor. This should be used over evaluating relative zones of increased TILs within the tumor.

4. All mononuclear cells (including lymphocytes and plasma cells) should be scored, but polymorphonuclear leukocytes (neutrophils) should be excluded.

Specific recommendations to melanoma included only assessing the vertical growth phase of the primary tumor. This novel system of reporting TIL in percentages will require validation to correlate with the older and established Clark and MIA grading scheme.

5. Future Directions

In an effort to standardize TIL description, Galon et al developed Immunoscore^®, an IHC and digital pathology-based assay that allows the quantification of two T-cell subsets (CD3⁺ and CD8⁺) in both the core and invasive margin of tumors [60]. Initially proposed for colorectal cancer, the prognostic value of Immunoscore^® has been shown to be associated with decreased disease recurrence and increased survival. Notably, this Immunoscore^® was more strongly associated with overall, disease-specific, and disease-free survival than TNM staging in colorectal cancer patients among stages I – III [61,62]. Similar immunotype classifications have since been described and associated with clinical outcomes in cervical cancer, clear cell renal carcinoma, breast cancer, esophageal adenocarcinoma, gastric cancer, and pancreatic ductal adenocarcinoma [63–68].

In summary, a variety of immunotype classification schemes have been described, which range from simple aggregates of multiple immune markers to more complex systems accounting for histologic patterns of immune infiltrates within the tumor itself. These classification paradigms allow for the simplification of complex data regarding spatial relationships between immune cells, intra-tumoral vasculature, and tertiary lymphoid structures, as well as relationships between multiple immunologic cell lines. Going forward, increasingly sophisticated classifications, digital TIL quantification, and further validation of these schemes will provide a more accurate framework for patient prognosis and therapeutic decision making.